[Haskell-cafe] bytestring vs. uvector

Roman Leshchinskiy rl at cse.unsw.edu.au
Mon Mar 9 00:37:02 EDT 2009


On 09/03/2009, at 11:47, Claus Reinke wrote:

> Btw, have any of the Haskell array optimization researchers
> considered fixpoints yet?

This, for instance, is a very nice paper:

http://www.pllab.riec.tohoku.ac.jp/~ohori/research/OhoriSasanoPOPL07.pdf

However, in the context of high-performance array programming explicit  
recursion is bad because it is very hard if not impossible to  
parallelise automatically except in fairly trivial cases. And if your  
array program is not parallelisable then you don't really care about  
performance all that much :-)

Roman




More information about the Haskell-Cafe mailing list